SEED - Self-learning and AI-powered Atmosphere and Energy Forecasting System using Earth Observation Data

You are here

Objectives of the service

The SEED solution offers a disruptive energy forecasting solution. Based on our innovative self-learning and AI-powered atmosphere forecasting system developed over the years powered by Satellite Earth Observation (EO) Data, our solution is able to forecast accurately energy production for wind and solar platforms.

Our solution will minimize uncertainty in energy production that arises due to uncertainties in atmospheric conditions and weather patterns by offering a software-based analytics engine that processes atmosphere, ocean EO data, energy production parameters as well as other related requirements. 

 

Users and their needs

Our targeted users are the renewable energy producers in India. It is due to the emerging needs of renewable energy production and ever-changing atmospheric conditions over a large geographical area, but also due to the increasing demand of energy production and its reliability for a rising population. Therefore, India is very attractive market for our proposed solution. SEED developments included a first stage version of software dashboard that presents atmospheric conditions, energy production status based on advanced models interlinked with weather parameters and innovative AI models, thus allowing for our service users in achieving the following:

  • Optimized power production planning (short-term, medium-term and long-term forecasting)
  • Reduced penalties related to imbalance costs (compliance with deviation settlement mechanism regulatory framework in India)
  • Optimized energy trading opportunities
  • Reduced maintenance operations/costs
  • Supporting grid operators in stabilizing the grid

Enable a faster transition towards a low carbon economy 

Service/ system concept

The concept is illustrated in the schematic below. The platform covers the following elements and features:

  • Access to satellite EO data products via API.
  • Data storage: Big Data platform (hardware and software) for storing updating EO data and sets of predictive models and forecasts per grid node.
  • Input parameters: Atmosphere and ocean EO data with spatial and temporal resolution depending on purpose and in-situ data (optional).
  • IT infrastructure: High-performance Computing (HPC) cluster for distributed, proprietary self-organising modelling and forecasting.
  • Predictive models: Up to ten thousands of self-organised predictive models reflecting 3D atmosphere columns of each to be forecasted input parameter.
  • Output parameters provided to customers: Probabilistic forecasts of a number of energy production data.
  • Forecast Horizon: From 15 minutes to 1 year.
  • Temporal Resolution: Minutes, hours, days, weeks, months.
  • Spatial resolution: Spatial resolution up to 10 km.
  • Spatial coverage: Global, with incidence on the Indian market/conditions.

Space Added Value

SEED forecasting solution is built using globally high-resolved space-based earth observation data of COPERNICUS missions combined with high performance computing and advanced AI algorithms. SEED is the first forecasting engine fully relying on non-interpolated EO data thus increasing the accuracy, validity, and reliability of the information used for modelling. SEED solution can accurately forecast several important energy parameters (and respective related weather and climate) parameters in short to medium-term horizon with high spatial resolution.

Current Status

SEED system has been implemented for a pilot solar energy generation customer in India. The implementation was successful, and it was shown that SEED forecasting was able to reduce the penalties to half in comparison to the available forecasting services. For a complete solution of the SEED platform for solar and wind energy with complete automation as per the required regulations of India the development of a market ready system is required.

Status Date

Updated: 24 August 2020 - Created: 24 August 2020